Agenda

• Using LLMs to your advantage:
  • RStudio GitHub Copilot
  • Ellmer (LLM Assistant in R)
• Setting up API keys for LLMs
• Integration tips & best practices

Introduction to GitHub Copilot in RStudio

• GitHub Copilot is an AI pair programmer helping to write code efficiently.
• Link for student sign-up:
  GitHub Education for Students
• Official RStudio Copilot docs:
  RStudio Copilot Setup

Installing & Enabling GitHub Copilot in RStudio

1. Confirm you have an active GitHub Copilot subscription.
2. Open RStudio.
3. Go to Tools > Global Options > Copilot tab.
4. Tick “Enable GitHub Copilot”
5. Click Install to download Copilot Agent components if prompted.
6. Click Sign In.
7. In the dialog, copy the Verification Code shown.
8. Paste the code on GitHub verification page to authenticate your RStudio IDE session.
9. Restart RStudio to finalize setup.

Using GitHub Copilot: Workflow & Features

• Copilot offers ghost text autocomplete suggestions as you type.
• Suggestions are AI-generated, not simple static completions.
• Hit Tab to accept; Arrow keys to cycle suggestions.
• Use Copilot to:
  • Write functions based on comments.
  • Generate boilerplate code.
  • Get context-aware suggestions.
• Example usage in R:

# Calculate the mean of a vector
mean_vec <- function(x) {
  # Copilot will suggest the function body here
}

How to video:

How to Ask Questions with GitHub Copilot

• Use a comment starting with # q: plus a question mark ? at the end.
• Copilot interprets this as a request for an answer and returns a comment response starting with # a:

Example:

# q: How to summarize data by group in dplyr?
# a: 
library(dplyr)
summary <- data %>%
  group_by(group_var) %>%
  summarize(mean_value = mean(target_var, na.rm = TRUE))

Getting Started with Other LLMs in RStudio

• Beyond Copilot, you can use Google Cloud-based LLMs (e.g., Gemini).
• First, create a Google Cloud Project.
• Then, generate an API key via Google Cloud Console.
• Store your API keys securely for package access.

What is an API Key?

• A unique token that gives your R session access to cloud services like language models.
• Must be kept private; never share in public code.
• Generate a API key with https://aistudio.google.com/app/apikey

Securely Storing Your API Keys in R

# Install 'usethis' if needed
# install.packages("usethis")

# Open .Renviron file to securely add environment variables
usethis::edit_r_environ()

Then add your key in the opened file:

GEMINI_API_KEY="your_api_key_here"

Save and restart RStudio to load the variables.

Introducing Ellmer: Your LLM Assistant in RStudio

• Ellmer makes it easy to access Google Gemini GPT-based chat from R.
• Install Ellmer package with dependencies:

install.packages("ellmer", dependencies = TRUE)

Starting an Ellmer Chat Session

library(ellmer)
# Create a Gemini chat session
chat <- ellmer::chat_google_gemini()

# Launch browser-based interactive chat interface
ellmer::live_browser(chat)

Using Ellmer to Boost Productivity

• Ask questions about your R code or data analysis.
• Request specific code snippets.
• Get explanations or suggestions for improving your scripts.
• Works as a complementary assistant alongside Copilot for a seamless workflow.

Best Practices for LLMs in RStudio

• Always review AI-generated code carefully before running.
• Combine Copilot and Ellmer to cover autocomplete and deeper Q&A.
• Protect your API keys by storing them in .Renviron.
• Keep your packages updated for best integration experience.
• Use LLMs to accelerate learning and coding — not to blindly copy code.

Summary & Next Steps

• Setting up GitHub Copilot in RStudio boosts your coding efficiency.
• Storing and using API keys securely is essential.
• Ellmer package wraps Google Gemini large language model chat inside RStudio.
• Try both tools for R programming and data analysis assistance.
• Practice regularly to build trust and understanding with AI-assisted coding.